# Worked Example Walkthrough

This walkthrough explains the lifecycle behind the fictional Maria E. Reyes example files. It is not a new schema requirement; it is a concrete way to read the examples as a small OCF workflow.

Files:

- `sample-resume-source.txt`: the source resume.
- `sample-resume.ocf.json`: the enriched OCF file.
- `sample-resume.md`: notes on the fictional review history.
- `sample-job-description.txt`: a fictional target job description for curation.

## 1. Source Resume Captured

The starting point is an ordinary short resume. In OCF terms, the resume is not the master truth. It is a `sourceArtifact`.

The importer or LLM should:

- record the resume as a `sourceArtifacts` entry;
- extract person, education, certifications, skills, experience, positions, and achievements;
- preserve useful wording from the resume;
- assign provenance to imported items;
- avoid inventing missing metrics, dates, supervisors, or outcomes.

At this point the file is best understood as an `imported-starter`, even if the example file later shows the richer candidate-master shape.

## 2. Intake Turns Bullets Into Career Memory

An OCF-oriented intake pass does more than rewrite resume bullets. It asks what each bullet means and whether it should become:

- a canonical achievement;
- structured metrics;
- supporting facts;
- a private reflection;
- a caution;
- an open question;
- a narrative variant for a particular audience.

The ransomware response story in `sample-resume.ocf.json` shows this pattern. The same underlying career event appears as:

- a structured achievement;
- supporting facts and metrics;
- private reflection material;
- cautions about overclaiming;
- narrative variants for different audiences.

That is the core OCF loop: preserve the facts once, then let future curation choose the right wording.

## 3. Review Adds Guardrails

Early drafts often overstate something. OCF records those corrections instead of making the user repeat them in every future session.

In the sample file, cautions and open questions show how a reviewer or LLM can preserve:

- claims not to make;
- missing attribution details;
- questions about how military leadership translates to civilian staffing contexts;
- areas where a future conversation should ask for better evidence.

Cautions are leading controls. A future curation pass should read them before drafting.

For example, the sample file records a caution against positioning Maria as an "AI / ML security specialist." The underlying fact is useful: she has operational exposure to ML-based detection tooling. The risky draft leap is turning that exposure into research-level AI/ML specialization. Once the caution is saved, a later tool can still use the real fact, but it should choose safer wording such as "evaluated and operationalized ML-assisted detection tooling" rather than overstating her expertise.

That is the value of a caution: the user corrects the mistake once, and future curation starts with that correction already in memory.

## 4. Candidate Master Accumulates Over Time

After review, accepted material can become part of a candidate-owned master OCF. The master may contain more than any resume should show:

- achievements that can be shared;
- private reflections that improve future coaching or interview prep;
- source artifacts;
- provenance;
- open questions;
- cautions;
- narrative variants;
- skills and certifications.

The master is not an output. It is the private memory layer.

## 5. Curation For A Target

When the user provides a target job description, a curator reads the master plus the target and decides:

- what to filter because of privacy or relevance;
- what to question because it is missing, stale, or unclear;
- what to rank because it is strong evidence for the target.

The output of this step is not necessarily a final resume. It may be:

- proposed improvements to the master;
- export-ready content for a resume or cover letter;
- interview-prep notes;
- a short list of questions the user should answer first.

If an item is real but irrelevant to the target, it stays in the master and gets filtered from the current output.

## 6. Export Produces Files

The exporter turns export-ready content into a file or paste-ready artifact:

- resume;
- cover letter;
- PDF;
- DOCX;
- HTML;
- LinkedIn-shaped paste bundle;
- JSON Resume;
- LER input;
- interview-prep packet.

The exporter should not decide which career claims belong. That judgment belongs to curation.

## 7. Updates Flow Back

After the output is drafted, the useful closing question is not just "does this resume look good?" It is:

> What did we learn that should be saved for next time?

Examples:

- a correction becomes a `caution`;
- a missing metric becomes an `openQuestions` item;
- a new story becomes a private reflection;
- a polished audience-specific bullet becomes a `narrativeVariant`;
- a confirmed claim becomes a canonical achievement update.

That feedback loop is why OCF gets more useful after each session.

## Minimal First-Time Flow

If a user has no OCF and only provides a resume and job description, a practical LLM flow is:

1. Ask briefly whether they already have an OCF.
2. Treat the resume and job description as source artifacts.
3. Build a provisional imported-starter view.
4. Ask the few questions that affect this target.
5. Produce the requested resume and cover letter.
6. Suggest the OCF updates worth saving.

That is enough. The user does not need to finish a complete master before getting value.